Comprehensive analysis of multi-omics vaccine response data using MOFA and Stabl algorithms.

Area of Science:

• Immunology
• Computational Biology
• Bioinformatics

Background:

• The FluPRINT dataset contains multi-omics data, including protein expression and cell counts, from individuals receiving a vaccine.
• Vaccine response is complex and influenced by various biological factors, necessitating advanced analytical approaches.

Purpose of the Study:

• To compare the performance of MOFA and Stabl algorithms in analyzing the FluPRINT multi-omics dataset.
• To identify key features and cell subsets that predict vaccine response using different analytical methods.
• To provide guidance on selecting appropriate analytical tools for large immunological multi-omics data.

Main Methods:

• The FluPRINT dataset was preprocessed, involving feature removal, assay-specific scaling, and outlier exclusion.
• Missing values were handled by removing features with high missingness and ignoring remaining missing data points.
• The MOFA and Stabl algorithms were applied to analyze population structure and predict vaccine response.

Main Results:

• MOFA identified IL neg 2 CD4 pos CD45Ra neg pSTAT5 as a key feature, explaining significant data variance with p < 0.05.
• Stabl identified CD33- CD3+ CD4+ CD25hiCD127low CD161+ CD45RA+ Tregs as the top predictive feature, aligning with previous findings.
• MOFA achieved an AUROC of 0.616, while Stabl achieved a slightly higher AUROC of 0.634 for predicting vaccine response.

Conclusions:

• Both MOFA and Stabl are valuable tools for analyzing complex immunological multi-omics data.
• Stabl demonstrated a slightly better ability to predict vaccine response compared to MOFA in this dataset.
• The study highlights the importance of algorithm selection for uncovering biological insights from multi-omics data.